CN-122024960-A - Intelligent adaptation selection system for industrial design materials and integrated application method for multi-scene structure device

CN122024960ACN 122024960 ACN122024960 ACN 122024960ACN-122024960-A

Abstract

The application provides an intelligent adaptation selection system for industrial design materials and an integrated application method for a multi-scene structure device, which are applied to the technical field of data processing. The method comprises the steps of carrying out full-flow intelligent processing around the model selection of an industrial design material, firstly carrying out standardized pretreatment and quantitative verification on design demand parameters, evaluating consistency through deviation coefficients to generate a standard data set, then realizing full-automatic or auxiliary material matching based on a deep learning matching model, outputting a candidate list and an adaptation score to construct a material-demand corresponding data set, improving XGBoost network by means of TensorFlow, improving precision through feature optimization and the like to generate adaptation ordering and feasibility probability, determining a final recommendation result through a multi-stage weight fusion strategy, and finally supporting interactive adjustment based on a visual platform to output a derivable model selection report and a parameter configuration file.

Inventors

SONG NING
ZHANG HONG
Di Guangbo
LIU DANYAN

Assignees

中聚创意设计产业(南京)有限公司

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (7)

1. An intelligent adaptation selection system for industrial design materials and an integrated application method for a multi-scene structure device are characterized by comprising the following steps: Preprocessing the industrial design demand parameters, uniformly converting the industrial design demand parameters into a standardized data format, quantifying key indexes, completing parameter matching verification by taking the product function requirement as a core reference and the scene use condition as a constraint condition, and evaluating the parameter consistency through a deviation coefficient to generate a standardized demand parameter data set; processing the standardized demand parameter data set based on the deep learning matching model, supporting full-automatic intelligent matching or auxiliary matching by a designer with a small amount of parameter adjustment, generating a candidate material list and an suitability score, extracting core performance indexes and scene adaptation key points, and synchronously associating material characteristic parameters to generate a material-demand corresponding data set; Processing the material-demand corresponding data set based on an improved XGBoost evaluation network under TensorFlow framework, improving the adaptation precision by adopting characteristic engineering optimization, sample equalization processing and multi-level loss function, and generating a material comprehensive proper matching degree sequencing result and a multi-dimensional feasibility probability value; Generating a final material adaptation recommendation result through early-stage functional weight distribution, medium-stage scene characteristic weighted fusion or later-stage dynamic weight adjustment voting/gradient lifting fusion strategy based on material performance; And processing the adaptation index, the performance parameter and the recommendation result based on the client visual interaction platform, supporting the designer to interactively modify the parameter and the scene constraint, and generating a exportable material model selection report and a parameter configuration file.
2. The method of claim 1, wherein preprocessing the industrial design requirement parameters, uniformly converting the industrial design requirement parameters into a standardized data format and quantifying key indexes, completing parameter matching verification by taking product function requirements as core references and scene use conditions as constraint conditions, evaluating parameter consistency by deviation coefficients, and generating a standardized requirement parameter data set, wherein the method comprises the following steps: Aiming at the multi-dimensional heterogeneous characteristics of the industrial design demand parameters including functional indexes, scene constraints and cost thresholds, the data format is regulated through format unified conversion, and then the quantitative treatment is carried out on the mechanical properties, weather resistance, processing feasibility, cost intervals and environmental protection levels, so that the quantitative standard and the precision range of each parameter are defined; Taking the core function requirement of the product as a core reference and the scene use condition as a rigid constraint, developing layering verification of basic parameter screening and core index adaptation, and checking the matching compliance of parameters, the function requirement and the scene constraint one by one; introducing a deviation coefficient evaluation mechanism to check parameter consistency, setting a parameter integrity threshold to perform double check, and establishing a grading processing mechanism of manual intervention of slight abnormality automatic correction and serious abnormality; through the whole flow process of format unification-index quantification-layering verification-consistency assessment-anomaly correction, a standardized demand parameter data set with data normalization, parameter integrity and material selection suitability is generated.
3. The method of claim 2, wherein processing the standardized demand parameter dataset based on the deep learning matching model, supporting full-automatic intelligent matching or designer small amount of parameter adjustment auxiliary matching, generating candidate material list and suitability score, extracting core performance index and scene adaptation key points and synchronizing associated material characteristic parameters, generating a material-demand correspondence dataset, comprising: Based on index dimension characteristics and material library characteristics of a standardized demand parameter data set, negotiating and determining an intelligent matching strategy by a deep learning matching model and an adaptation rule library, and determining a matching retrieval range and a similarity threshold value by demand parameter priority analysis and material characteristic association statistics; Splitting the standardized demand parameter data set by adopting a parameter classification retrieval mechanism, extracting core parameters according to the function index, the scene constraint and the cost threshold type, and associating the corresponding material characteristic matching labels to generate a preliminary candidate material set; Performing suitability verification on the preliminary candidate material set, starting a supplementary search service aiming at materials with performance parameters which do not reach the standard, triggering a rule adjustment mechanism on scene adaptation conflict materials, executing a priority rearrangement process on samples with cost exceeding a threshold value, and generating an optimized candidate material list containing a material screening strategy and an adaptation correction scheme; carrying out suitability scoring accounting on the optimized candidate material list, and synchronously generating a material suitability scoring report and characteristic association details by quantifying the function matching degree, the scene matching degree and the cost controllability by combining a multidimensional weighting algorithm; and (3) supporting full-automatic intelligent matching or auxiliary matching by a designer with a small amount of parameter adjustment, extracting core performance indexes and scene adaptation key points, and synchronously associating material characteristic parameters to form a material-demand corresponding data set finally containing material information, suitability scores and characteristic association data.
4. The method of claim 1, wherein processing the material-requirement correspondence dataset based on the improved XGBoost evaluation network under the TensorFlow framework, using feature engineering optimization, sample equalization processing, and multi-level loss function to improve adaptation accuracy, generating a material ensemble fitness ranking result and a multi-dimensional feasibility probability value, comprises: Processing the adaptation feature complexity of the material-requirement corresponding dataset, definitely comprising a function matching precision threshold and core evaluation indexes of scene constraint fit conditions, classifying and labeling, and quantitatively evaluating TensorFlow framework computing resources to generate feature complexity classification results and computing resource evaluation data; Determining a fine or high-efficiency training strategy by a supply-demand balance algorithm based on the feature complexity classification result and the computing power resource evaluation data, and defining a model input data range by taking the core feature full coverage as a principle to generate a model training strategy and an input data range definition result; Performing feature engineering optimization and sample equalization processing on the data set according to the input data range definition result to generate an optimized material-requirement corresponding data set; Constructing a multi-level loss function, importing an optimized material-demand corresponding data set into an improved XGBoost evaluation network, determining iteration times and learning rate adjustment rhythm by combining an upper limit of calculation force, and generating a configured model training scheme; And performing iterative training according to a model training scheme, outputting the multidimensional adaptive probability of the material, calculating the comprehensive score through weighted fusion, and generating a comprehensive adaptive degree sequencing result and a multidimensional feasibility probability value of the material.
5. The method of claim 4, wherein generating a final material fit recommendation by pre-functional weight assignment, mid-scene characteristic weighted fusion, or post-dynamic weight adjustment voting/gradient boost fusion strategy based on material performance comprises: Performing demand matching degree verification and weight rationality evaluation on the functional index priority, the weight quantization standard and the adaptation association rule in the early-stage functional weight allocation strategy to generate functional weight allocation coefficients, core index adaptation threshold values and weight adjustment constraint conditions, so as to form early-stage fusion core mechanism information; Scene dimension disassembly, characteristic weight distribution and fusion logic adaptation in the middle-stage scene characteristic weighted fusion are performed with scene suitability verification and fusion efficiency measurement and calculation, scene characteristic adaptation parameters, dynamic weighted adjustment coefficients and scene-material matching association schemes are generated, and middle-stage fusion basic information is formed; Combining the performance weight calibration, result coordination rules and probability integration flow in later-stage dynamic weight adjustment voting/gradient lifting fusion, performing strategy coordination optimization and redundant information elimination processing on the early-stage fusion core mechanism information and the middle-stage fusion basic information, generating multi-strategy linkage execution parameters and result calibration correction coefficients, and forming fusion optimization information; And integrating the early-stage fusion core mechanism information, the middle-stage fusion basic information and the fusion optimization information to generate a final material adaptation recommendation result considering the function suitability, the scene fitness and the performance stability.
6. An electronic device, comprising: and a memory for storing executable instructions of the first processor; The first processor is configured to execute the integrated application method of the intelligent adaptation type selection system of industrial design materials and the multi-scene structure device according to any one of claims 1-5 by executing the executable instructions.
7. A computer readable storage medium having a computer program stored thereon, wherein the computer program when executed by a second processor implements the integrated application method in an intelligent adaptation selection system for industrial design materials and a multi-scene structure device according to any one of claims 1 to 5.

Description

Intelligent adaptation selection system for industrial design materials and integrated application method for multi-scene structure device Technical Field The invention relates to the technical field of data processing, in particular to an intelligent adaptation and selection system for industrial design materials and an integrated application method for a multi-scene structure device. Background Firstly, the demand parameters lack standardized processing, the parameter formats under different design scenes are not uniform, and key indexes are not quantized enough, so that the accuracy of parameter matching verification is low, and material and demand adaptation deviation is easy to occur; the method has the advantages that the method is simple in structure, convenient to use, and easy to operate, the model selection process is highly dependent on experience of a designer, has the cognitive bias errors such as availability heuristic bias and anchoring effect, the decision error rate is up to 42%, is difficult to cope with the dimension problem of multi-objective optimization, cannot balance multi-scene constraint conditions efficiently, and is difficult to support a designer to efficiently finish parameter adjustment and scheme confirmation due to the fact that the existing intelligent model selection technology has the problems of insufficient data quality, high algorithm complexity, low calculation efficiency and the like, systematic application of feature engineering optimization and sample equalization processing is lacking, the adaptation precision is difficult to meet high-end manufacturing requirements, and the model selection result is lack of a dynamic adjustment mechanism, real-time weight distribution cannot be optimized according to scene characteristic change or material performance. The current industrial situations of aggravation of the risk of a key material supply chain, reinforcement of environmental protection policy constraint and the like put higher requirements on the scientificity and the high efficiency of material selection. The traditional trial-and-error method and the empirical method have the defects of high time cost, serious resource waste, poor adaptation consistency and the like, and can not meet the requirements of modern industrial design on quick response, accurate matching and low-cost model selection. It should be noted that the information disclosed in the above background section is only for enhancing understanding of the background of the present disclosure and thus may include information that does not constitute prior art known to those of ordinary skill in the art. Disclosure of Invention Other features and advantages of the application will be apparent from the following detailed description, or may be learned by the practice of the application. According to one aspect of the application, an intelligent adaptation type selection system of industrial design materials and an integrated application method of a multi-scene structure device are provided, which comprise the steps of preprocessing industrial design requirement parameters, uniformly converting the industrial design requirement parameters into a standardized data format and quantifying key indexes, completing parameter matching verification by taking product function requirements as a core reference and scene use conditions as constraint conditions, evaluating parameter consistency through deviation coefficients to generate a standardized requirement parameter data set, processing the standardized requirement parameter data set based on a deep learning matching model, supporting full-automatic intelligent matching or designer small quantity parameter adjustment auxiliary matching to generate a candidate material list and an adaptation score, extracting core performance indexes and scene adaptation key points and synchronously associating material characteristic parameters to generate a material-requirement corresponding data set, processing the material-requirement corresponding data set based on an improved XGBoost evaluation network under a TensorFlow framework, adopting characteristic engineering optimization, sample equalization processing and multi-level loss function to promote adaptation accuracy, generating a material comprehensive adaptation degree sequencing result and a multi-dimensional feasibility probability value, carrying out front-stage function weight distribution, weighting or later stage weighting based on material comprehensive performance, supporting performance adjustment based on a dynamic performance of a dynamic platform, generating a material adaptation performance gradient, and a final interaction type recommendation parameter, and leading out interaction type recommendation and a client-stage performance parameter adjustment based on the improved performance index. According to still another aspect of the application, an electronic device comprises a first processor and a memo